A library designed to ingest and pre-process messy documents like PDFs and HTML for use in LLM training.